Suspend to disk - why does it hurt so?
Over the past few years, suspend to RAM support on Linux has improved by
leaps and bounds. Sadly, the same can not really be said about suspend
to disk. Two competing kernel implementations exist with varying levels
of features, but both follow the same basic design concept. Arguments
are made about details and little overall progress is made.
This presentation will cover the design concepts used for the existing
implementations, along with their shortcomings and why fundamental
issues may prevent them from ever becoming entirely reliable. New
concepts in suspend to disk will be presented, such as utilising the
kexec framework or implementing a checkpoint-like system with a focus on
saving application state rather than the entirety of kernel state.
This presentation will assume some familiarity with the kernel and unix
architecture, but no deep technical knowledge will be required.
Project: Linux kernel and suspend to disk
Matthew Garrett
Matthew Garrett takes a break from the inherent complexity of fruitfly
biology by working on the clean, simple and straightforward ACPI
specification. Head of the Ubuntu laptop team, Matthew has gained
extensive experience of laptops, power management and the myriad ways
that they can go horribly wrong.
Matthew is always interested in meeting representatives of laptop
manufacturers and asking them what on earth they were thinking.
Matthew Garrett
Matthew Garrett takes a break from the inherent complexity of fruitfly
biology by working on the clean, simple and straightforward ACPI
specification. Head of the Ubuntu laptop team, Matthew has gained
extensive experience of laptops, power management and the myriad ways
that they can go horribly wrong.
Matthew is always interested in meeting representatives of laptop
manufacturers and asking them what on earth they were thinking.
